Neural Network Multiphase Fluid Flow Equilibrium Modeling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for modeling multiphase fluid flow in porous media are computationally intensive and time-consuming, particularly due to the need for frequent thermodynamic and geochemical equilibrium calculations, which dominate the simulation time and resource requirements.

Innovation Solution

Implementing neural networks to determine thermodynamic and geochemical equilibria, replacing traditional equation-solving methods, allowing for faster and more efficient calculation of fluid properties and compositions at equilibrium.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional equation-solving methods are used to determine thermodynamic and geochemical equilibria, then accuracy in modeling fluid properties is maintained, but computational time and resources are significantly increased

Engineering Contradiction:
Improveaccuracy in modeling fluid propertiesVSAvoidcomputational time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-training neural networks offline using extensive thermodynamic and geochemical equilibrium data. This training phase is performed beforehand, allowing the trained models to rapidly predict equilibrium properties during actual simulations without performing computationally intensive calculations at runtime. The neural networks are prepared in advance to handle the complex equilibrium determinations efficiently during production use.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent substitutes the traditional mechanical equation-solving approach with a neural network-based predictive system. Instead of iteratively solving complex thermodynamic and geochemical equilibrium equations during simulation, the system uses pre-trained neural networks that have learned the equilibrium relationships from training data, replacing the computational mechanics with a faster artificial intelligence-based prediction mechanism.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If traditional equation-solving methods are used to determine thermodynamic and geochemical equilibria, then accurate fluid composition calculations are achieved, but computational resources and simulation time are significantly increased

Engineering Contradiction:
Improveaccuracy in fluid composition calculationsVSAvoidsimulation efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies preliminary action by pre-training neural networks offline using extensive thermodynamic and geochemical equilibrium data. This training phase is performed beforehand, allowing the trained models to rapidly predict equilibrium properties during actual simulations without performing computationally intensive calculations at runtime. The neural networks are prepared in advance to handle the complex equilibrium determinations efficiently during production use.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent substitutes the traditional mechanical equation-solving approach with a neural network-based predictive system. Instead of iteratively solving complex thermodynamic and geochemical equilibrium equations during simulation, the system uses pre-trained neural networks that have learned the equilibrium relationships from training data, replacing the computational mechanics with a faster artificial intelligence-based prediction mechanism.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Reliability

If complex equilibrium equations are solved at each time step and mesh, then accurate multiphase fluid flow modeling is achieved, but computational complexity and time requirements increase significantly

Engineering Contradiction:
Improveaccuracy in multiphase fluid flow modelingVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-training neural networks offline using extensive thermodynamic and geochemical equilibrium data. This training phase is performed beforehand, allowing the trained models to rapidly predict equilibrium properties during actual simulations without performing computationally intensive calculations at runtime. The neural networks are prepared in advance to handle the complex equilibrium determinations efficiently during production use.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent substitutes the traditional mechanical equation-solving approach with a neural network-based predictive system. Instead of iteratively solving complex thermodynamic and geochemical equilibrium equations during simulation, the system uses pre-trained neural networks that have learned the equilibrium relationships from training data, replacing the computational mechanics with a faster artificial intelligence-based prediction mechanism.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP2791712B1Method and system for dynamically modeling a multiphase fluid flow
Publication Date: 2016.03.09 TERRA 3E
  • EP2791712B1 patent drawingFigure 1~2
  • EP2791712B1 patent drawingFigure 3
  • EP2791712B1 patent drawingFigure 4

AI summary

The invention relates to a method for dynamically determining at least one physicochemical property and the composition of a multiphase fluid flow in a porous medium, including: constructing a geological model of said porous medium and discretizing the model into basic cells; determining (40) an initial status of each cell, including determining the composition of said cell; and determining (42) a change, in at least one cell and during at least one predetermined time interval, in at least one physicochemical property and in the phase composition of said fluid in each component, comprising at least one step (52, 54) of determining an equilibrium within said cell, including the use of at least one neuronal network to determine characteristics that are in equilibrium in said fluid from the characteristics thereof that are out of equilibrium.